Targeting the FtsZ Allosteric Binding Site with a Novel Fluorescence Polarization Screen, Cytological and Structural Approaches for Antibacterial Discovery

Bacterial resistance to antibiotics makes previously manageable infections again disabling and lethal, highlighting the need for new antibacterial strategies. In this regard, inhibition of the bacterial division process by targeting key protein FtsZ has been recognized as an attractive approach for discovering new antibiotics. Binding of small molecules to the cleft between the N-terminal guanosine triphosphate (GTP)-binding and the C-terminal subdomains allosterically impairs the FtsZ function, eventually inhibiting bacterial division. Nonetheless, the lack of appropriate chemical tools to develop a binding screen against this site has hampered the discovery of FtsZ antibacterial inhibitors. Herein, we describe the first competitive binding assay to identify FtsZ allosteric ligands interacting with the interdomain cleft, based on the use of specific high-affinity fluorescent probes. This novel assay, together with phenotypic profiling and X-ray crystallographic insights, enables the identification and characterization of FtsZ inhibitors of bacterial division aiming at the discovery of more effective antibacterials.


■ INTRODUCTION
New antibiotics are urgently needed to cope with the global rise of bacterial pathogens resistant to antibiotics in use, which renders lethal infections that once were treatable and controllable. 1−4 To discover new antibiotics, processes essential for bacterial reproduction and spreading must be targeted, such as bacterial cell division, a clinically unexplored target. While eukaryotic mitosis has been vastly exploited for cancer treatments, targeting the bacterial cell division remains a clinical challenge. In this context, considerable knowledge has been gathered on the function, structure, dynamics, and interacting partners of essential cell division protein, FtsZ, since its localization forming a ring at the division site was discovered. 5 The FtsZ ring orchestrates the assembly of the divisomal machinery in most bacteria. 6 It is formed by FtsZ clusters that undergo treadmilling and guide cell envelope invagination at the division site. 7−9 Bacterial division and FtsZ have been recognized as attractive targets for discovering new antibiotics. 10−12 However, for many inhibitors reported in the literature, FtsZ targeting has not been really evidenced, slowing the progress in FtsZ-inhibitor development. 13 FtsZ was validated as an antibacterial target of the potent antistaphylococcal experimental inhibitor PC190723 (Figure 1), 14 which is synergistic with β-lactams. 2,15 However, in spite of consid-erable synthetic efforts, the benzamide derivative PC190723, many analogues, and other inhibitors, 16 have not made it into the clinic, possibly due to unsuitable pharmacological properties and a relatively high frequency of resistance mutations. Only one prodrug analogue of PC190723, TXA709 17 (Chart S1), was designated as a qualified infectious disease product for Staphylococcus aureus infections and has recently completed a phase I clinical trial. 18 FtsZ filaments form by head to tail self-association of FtsZ monomers; this is followed by hydrolysis of FtsZ-bound guanosine triphosphate (GTP) that triggers monomer dissociation. Cooperative assembly of single-stranded FtsZ filaments is made possible by FtsZ monomers switching between a low and a high association affinity state during the FtsZ filament assembly. 19−21 In relaxed unassembled FtsZ monomers from several species, a side cleft between the Nterminal GTP-binding and the C-terminal subdomains is closed, whereas this cleft opens in tense FtsZ monomers from S. aureus (SaFtsZ) forming crystal filaments, 22 providing a binding site for allosteric modulators of FtsZ assembly. Indeed, PC190723 acts as an allosteric inhibitor that stabilizes FtsZ filaments against disassembly upon GTP hydrolysis and selectively inhibits cell division in Bacillus and Staphylococcus species. 23,24 PC190723 fits snugly in a flat conformation into its binding site at the bottom part of the open cleft in SaFtsZ crystals, 15,25 leaving apparently little possibility for chemical modification. However, there is an unfilled pocket above connected by a tunnel to PC190723 (encircled in Figure 1A,B) and also a broader cavity around this pocket, formed by helices H6 and H7 and the C-terminal subdomain. These cavities are known since the first FtsZ structure was described 26 but lack cognate ligands. Thus, there appears to be room for other ligands binding into the interdomain cleft in addition to PC190723.
We developed fluorescent probes for the interdomain cleft of FtsZ, in which a small fluorophore replaces the chlorothiazolopyridine tail of inhibitor PC190723. Among them, 4chloro-7-nitro-2,1,3-benzoxadiazole (NBD) probe 1 ( Figure  1C) was employed to demonstrate the FtsZ assembly switch in  solution, in which the side cleft between the N-and C-terminal domains, containing the benzamide binding site, opens in assembled FtsZ subunits and closes in unassembled monomers. 27 The missing crystal structures of closed, relaxed SaFtsZ monomers were then determined, 28,29 supporting an equilibrium between the open and closed-cleft conformers, relaxing to closed by the dissociation of tight longitudinal contacts in SaFtsZ filaments. 30 The FtsZ polymerization-induced assembly switch enables the filament treadmilling displacement 29  In this work, we describe the first competitive binding assay for screening FtsZ ligands based on the use of new high-affinity fluorescent probes 2 and 5 ( Figure 1C). Complementary phenotypic assays and X-ray crystallography have allowed the validation of this screen for the identification of FtsZ allosteric inhibitors. Hence, a series of structurally simplified benzamide analogues 6−26 ( Figure 2) have been synthesized and evaluated in this new assay, identifying FtsZ-targeting inhibitors of bacterial division. Altogether, the biological and structural studies reported in this work provide relevant information about the molecular recognition of FtsZ allosteric ligands, which may contribute to the discovery of more effective antibacterial agents.

■ CHEMISTRY
The synthesis of fluorescent probes 2−5 is outlined in Scheme 1. Racemic probes 2−4 were prepared starting from 2,6difluoro-3-hydroxybenzamide (27), which was O-alkylated with the appropriate N-Boc-2-bromoalkylamine 28−30, followed by the removal of the Boc group by treatment with trifluoroacetic acid and coupling of the resulting intermediates 35−37 with NBD chloride. For the synthesis of (R)-and (S)-2, the Mitsunobu reaction under microwave (MW) irradiation between benzamide 27 and the corresponding enantiomeric form of N-Boc-aminoalcohol 31, subsequent deprotection of the amino group, and reaction with NBD chloride afforded the desired enantioenriched probes. Racemic probe 5 and its enantiomers were obtained by applying the same sequence of Mitsunobu reaction, deprotection, and NBD-coupling but using the appropriate form of primary alcohol 38.
Benzamide bioisosteres 15−17 were synthesized as depicted in Scheme 3. The reaction of 3-[(3-chlorobenzyl)oxy]benzaldehyde with trimethyl(trifluoromethyl)silane afforded trifluoromethyl alcohol 41, which was oxidized to ketone 42 with the Dess−Martin reagent, followed by reductive amination using tert-butanesulfinamide as an ammonia equivalent to provide target compound 15. Imidazole derivatives 16 and 17 were prepared by the Pd-and Cumediated cross-coupling reaction of imidazole and 3iodoanisole, subsequent demethylation with boron tribromide of 43, and O-benzylation of phenol 44 with the corresponding chloro-or trifluoromethyl-substituted benzyl bromide.

■ RESULTS AND DISCUSSION
Enhanced Affinity Fluorescent Probes for the Interdomain Cleft of FtsZ. A key step in the development of a fluorescence-based binding assay targeting the FtsZ interdomain cleft is the generation of specific probes for this site. These tools should undergo a sensitive modification of their fluorescence properties when bound to the protein, which returns to those of the free probe when displaced by a reference ligand such as PC190723. Competition assays of this type require that the probe and the competitor have commensurate affinities, resulting in a limited measurement range of competitor affinities with a single probe. 33 Thus, specific medium affinity probes are required for the effective detection of weak but specifically competing molecules during inhibitor screening, whereas high-affinity probes are needed for the measurement of the dissociation constant (K D ) of highaffinity compounds. Our previous studies demonstrated that probe 1 ( Figure 1C) and its analogues with longer spacers between the difluorobenzamide ring and the fluorophore specifically bind in a narrow affinity range (K D = 11−26 μM) with moderate fluorescence anisotropy changes 27 that were deemed insufficient for the development of a competitive assay. However, their biochemical effects, together with molecular dynamics simulations of model complexes, supported the notion that an extension of the PC190723 binding site is available for inhibitors binding into the FtsZ interdomain cleft. 27 Thus, taking probe 1 as the starting  (Table S1), acquired employing the probe fluorescence. (C) Displacement curves of (S)-2 (3 μM) from stabilized FtsZ-GMPCPP polymers (8 μM binding sites) by PC190723 (black circles, K D < 0.1 μM), and its moieties 2,6difluoro-3-methoxybenzamide (DFMBA, K D = 0.67 ± 0.05 mM) (red), and 6-(chloro [1,3]thiazolo [5,4-b]pyridin-2-yl)methanol (CTPM, K D = 1.0 ± 0.1 mM) (blue). Solid lines are best fits to the data of a single site competition model (Experimental Section). The addition of guanosine diphosphate (GDP) (void circles) does not displace (S)-2 from its binding site. point, we resorted to adding aliphatic side chains to the spacer to gain additional hydrophobic interactions by increasing the length of the aliphatic chain according to the space available in the allosteric interdomain cleft. These new NBD-based compounds 2−5 ( Figure 1C and Scheme 1) were characterized by their fluorescence spectra ( Figure S1), the anisotropy changes upon specific reversible binding to FtsZ polymers ( Figure 3A), fluorescence microscopy of Bacillus subtilis cells ( Figure 3B), and binding affinity titrations ( Figure  S2 and Table S1). FtsZ polymers were formed in the presence of the slowly hydrolyzable GTP analogue GMPCPP and magnesium. The four probes showed large anisotropy values in the presence of FtsZ polymers compared to the free probe ( Figure 3A). However, the recovery of the initial anisotropy values after the addition of excess competing PC190723, supporting binding specificity, was only achieved in the case of probes 1, 2, and 5. Compounds 3 and 4, with the longer side chains, exhibited an increase in background anisotropy with unassembled FtsZ in the absence of magnesium that were deemed nonspecific. In addition, the anisotropy of 4 with FtsZ polymers in the presence of magnesium was not reduced by PC190723 competition ( Figure 3A). We thus focused on  Table S1) and a large anisotropy change compared to 1, whereas (R)-2 had a very small fluorescence anisotropy increase ( Figure 3A). Better results were obtained when the propyl side chain was incorporated at the carbon next to the amino group, and (R)-5 resulted in the highest affinity probe (K D = 1.9 ± 0.6 μM) and broadest anisotropy changes, whereas (S)-5 presented lower affinity (K D = 15 ± 2 μM) (Table S1 and Figure 3A). The replacement of the NBD by boron dipyrromethene (BODIPY) or acrylodan fluorophores, as well as the substitution of the oxygen of the alkoxybenzamide by nitrogen, resulted in smaller or nonspecific anisotropy changes (probes SP1−SP5 Figure S1, synthesis in Scheme S1). Treating B. subtilis cells with probes 2 and 5 ( Figure 3B) revealed that (S)-2 and both enantiomers of 5 (at 25−100 μM) induced the characteristic filamentous phenotype due to the inhibition of cell division, supporting FtsZ targeting. 31,32 These probes faintly stained intracellular structures, similarly to 1, 27 whereas (R)-2 was inactive in these assays. Interestingly, the higher affinity probe (R)-5 showed the best efficacy in the impairment of cell division and displayed antibacterial activity in B. subtilis (Table S1). Altogether, cytological profiling and in vitro binding measurements were in qualitative agreement and support the suitability of fluorescent probes (S)-2 and (R)-5 for the development of a competitive binding assay.
Competitive Fluorescent Method for Ligand Binding into the FtsZ Interdomain Cleft. Capitalizing on the anisotropy decrease of the fluorescent benzamide probes (S)-2 and (R)-5 upon dissociation, we designed homogeneous assays to specifically determine the binding of any molecule to the PC190723 binding site in FtsZ polymers. It should be noted that this assay is more challenging than the ones described for ligands binding into the nucleotide orthosteric binding site of FtsZ monomers. 34,35 In this new method, stabilized FtsZ polymers are required that do not disassemble upon nucleotide hydrolysis or by the action of other FtsZ polymerization inhibitors binding elsewhere in the protein molecule, producing misleading anisotropy changes. Gentle cross-linking of the FtsZ-GMPCPP polymers with 0.15% (v/v) glutaraldehyde, adapted from microtubule-fluorescent taxoid studies, 36 rendered stabilized FtsZ from B. subtilis (BsFtsZ) polymer preparations that were resistant to depolymerization by GDP addition but still capable of probe binding with the same affinity values as noncross-linked polymers ( Figure S2 and Table S1). We then employed mixtures of stabilized FtsZ polymers and probe (S)-2 in competition assays, measuring the decrease in fluorescence anisotropy upon displacement of the probe to determine binding at increasing concentrations of competing test ligands. As a proof of concept, Figure 3C shows the competition binding isotherms of the reference highaffinity ligand PC190723 and its weakly binding moieties DFMBA and CTPM. Importantly, when the potent polymerization inhibitor GDP was used as a negative control, no change in probe anisotropy was observed.
Binding Screen for Allosteric FtsZ Inhibitors Combined with Cytological Methods. Once the fluorescence method for ligand binding into the interdomain cleft was set up, we evaluated its potential to screen for FtsZ allosteric inhibitors. Complementary to this screen, cytological profiling tests of the identified binders allow ascertaining their ability to target FtsZ in bacterial cells. Fluorescence anisotropy measurements of probe (S)-2 at two concentrations of each tested compound (20 and 200 μM) were acquired; the solubility range of positively testing compounds was determined spectrophotometrically. Of note, the screen results can be displayed as raw anisotropy data, permitting to distinguish at a glance weak inhibitors from higher affinity candidates that markedly decrease the anisotropy values at both concentrations ( Figure 4A). Among several previously described FtsZ inhibitors, two selective disruptors with an unknown binding site, PC170942 35,37 and zantrin Z3, 38,39 gave no evidence of binding in our assay, and the specific GTP-replacing FtsZinhibitor UCM44 35 showed a marginal inhibitory effect (chemical structures in Chart S2). The promiscuous inhibitors resveratrol, 40 plumbagin, 41 and tiplaxtinin 42 (Chart S2) weakly modified anisotropy and total fluorescence intensity, although none of them impaired B. subtilis cell division or GTP-FtsZ ring localization in our hands. It should be noticed that aggregating compounds may sequester the protein or the probe or may scatter highly polarized light, potentially interfering with the fluorescence anisotropy measurements. Colored and fluorescent compounds can also interfere. Compounds initially testing positive in the anisotropy screen should thus be checked for any unexpected changes of the total fluorescence intensity in the same samples, as well as for any artefactual effects on replicates omitting FtsZ.
Next, N-or O-(alkylphenyl)benzamides 6−14, 18−26, and amide bioisosteres 15−17 were evaluated as FtsZ allosteric inhibitors (Figures 2 and 4A,B). Different types of structural modifications around the three main structural features of benzamide inhibitors 16 were explored: (i) linear or branched alkylenamino or alkylenoxy linkers of different lengths as spacers, similar to the recent structure−activity relationship studies; 43 (ii) difluorination of the benzamide core; and (iii) the position and effects of substituents around the phenyl ring selected as the aromatic moiety. Starting with lower affinity compounds, 3-and 4-aminobenzamides 6, 7, 9, and 10 were identified as weak binders ( Figure 4A). Complementary competition measurements indicated K D values around 0.2 mM in an affinity order of 7 > 9 ≥ 6 ≥ 10 ( Figure S3). Interestingly, compound 7 with a methylamino linker, markedly inhibited B. subtilis cell division at 0.75 mM ( Figure  4C), inducing a characteristic filamentous phenotype ( Figure  4D), whereas its propyl-amino analogue 8 was toxic to the bacterial cells. In addition, compound 7 (0.2 mM) effectively induced irregularly spaced Z-rings and delocalization of GFP-FtsZ into punctate foci ( Figure 4E), characteristic of alkoxybenzamide inhibitors. 14,44 These results indicate that our fluorescence binding screen method combined with phenotypic tests can effectively detect relatively weak FtsZ-targeting compounds, which may lead to more effective cell division inhibitors. Note that 7 inhibits B. subtilis division at a ∼20-fold lower concentration than the weak inhibitor 3-methoxybenzamide, 44 which was the starting point in the PC190723 development. 14 The other way around, compounds inhibiting bacterial division may now be tested in fluorescence anisotropy assays for potentially targeting the allosteric interdomain cleft (this work) or the nucleotidebinding site of FtsZ. 34,35 Enhancing the Affinity of Simplified Benzamide Inhibitors. Evaluation of 3-alkoxybenzamide 11 45 showed that this oxygen derivative was more effective than nitrogen analogues 6−10 in our anisotropy screen ( Figure 4A).

Journal of Medicinal Chemistry
pubs.acs.org/jmc Article Compound 11 had more than 10-fold lower K D (10 μM, Figure 5A) and a minimal cell division inhibitory concentration (MDC) value (Table 1) than its 3-aminobenzamide analogue 6. The K D values, determined with the high-affinity probe (R)-5, are useful to compare the affinities of the different inhibitors among them and correlate with the corresponding MDC values. Derivatives of 11, bearing an ethoxy or a propoxy spacer (12 or 13, respectively), had similar screen results ( Figure 4A) and cell division inhibitory activities (Table 1), whereas the 4-substituted analogue 14 was toxic. Replacement of the amide group of 11 by isosteres gave compounds 15−17 with reduced inhibition ( Figure 4A) that were also toxic to the cells. However, the addition of fluorine atoms at 2-and 6positions of the benzamide ring of 11 (as in PC190723) rendered the 3-chloro derivative 18 45,46 and the 3-trifluoromethyl analogue 20 that were both stronger competitors ( Figure 4A) with 1.3 and 0.7 μM K D values, respectively ( Figure 5A and Table 1). Compound 20 induced a marked divisomal inhibition phenotype on B. subtilis, consisting of cell filamentation and GFP-FtsZ delocalization ( Figure 5B and Table 1; raw data plots in Figure S4 and statistics in Table S2), as well as characteristic enlarged spherical cells in S. aureus and FtsZ-mCherry delocalization ( Figure 5C). Analogues of 20 with ethoxy and propoxy linkers 22 and 23, respectively, were weaker binders, as were the 2-and 4-trifluoromethyl analogues 19 and 21. Adding a positively charged dimethylaminoethyl side chain at the linker rendered analogue 24 water-soluble, although canceled binding and cellular activity ( Figure 4A and Table 1). At this point, we sought to further enhance inhibitor affinity with an analogous strategy to that employed with the fluorescent probes above. The starting point was compound 25 with an aminoethoxy linker between the two aromatic subunits, which is a higher affinity 3-trifluoromethylphenyl equivalent of the NBD probe 1 with comparable K D and cellular activity values to 20 ( Figure 5 and Table 1). The incorporation of a propyl side chain to the linker in (R)-26, equivalent to probe (R)-5, resulted in maximal affinity in the series (K D = 0.2 μM) with a 1 μM division inhibitory activity on B. subtilis cells, comparable to PC190723 ( Figure 5B and Table 1; raw data plots in Figure S4 and statistics in Table S2). The (S)-26 enantiomer has 10−20-fold weaker affinity and MDC than the (R)-enantiomer, which can be related to the 8fold reduced affinity of probe (S)-5 with respect to (R)-5.  (Table 1), as may be expected from the conservation of the PC109723 binding site in these two species.
Compound 20 is active against methicillin-resistant S. aureus Mu50, S. haemolyticus, and S. epidermidis (MIC 25−50 μM). However, in contrast to PC190723, the MDC value of 20 on S. aureus Mu50 is 5-fold higher than on B. subtilis, suggesting that the 3-trifluoromethylphenyl tail confers somewhat less activity against this strain than the chloro-thiazolopyridine tail. Then, the S. aureus MIC marginally decreases from 20 to (R)-26, compared to a 10-fold decrease in B. subtilis, indicating that the linker modifications that enhance affinity improve the  (Table S4). Most mutations mapped to the f tsZ gene, at positions corresponding to amino acids (P115), (V151), A182, G196, V214, A237, L249, L261, M262, N263, (A285), V297, T309, and (T358). These residues predominantly cluster around the interdomain cleft (exceptions in parenthesis) and concentrate at the PC190723 binding site ( Figure 5D); several of them are typically modified in PC190723-resistant mutants 14 (in the bold type above). Therefore, the resistance mutation results strongly support that the inhibitors tested target S. aureus FtsZ at the interdomain cleft. The reduced activity in comparison to B. subtilis might be caused by permeability differences or by the action of staphylococcal multidrug resistance efflux pumps.

Structural Insights into Inhibitor Binding to S. aureus
FtsZ. To confirm the proposed interaction of the inhibitors identified in our fluorescence screening, as a validation of the ligand-binding assay, we were able to determine the crystal structures of the SaFtsZ folded core (residues 12−316) 34 in complex with inhibitors 18, 20, and DFMBA to 1.55−1.70 Å resolution ( Figure 6 and Table S5). Omit electron density maps around the compounds clearly prove that the three molecules bind into the interdomain cleft ( Figure S5), as PC190723, 15,25 TXA6101, and TXA707 47 (chemical structures in Chart S1). The protein structure in the three complexes is essentially the same as in inhibitor-free SaFtsZ (PDB 3VOA, mean root-mean-square deviation (rmsd) = 0.39 Å), the SaFtsZ-PC190723 complex (PDB 4DXD, mean rmsd = 0.41 Å), and the SaFtsZ-TXA6101 complex (PDB 5XDU, mean rmsd = 0.43 Å). Compounds 18 and 20 slightly increase the distance between helix H7 and the C-terminal domain by 0.5 Å as compared to DFMBA, while a larger displacement of 1 Å is observed in the other cases.
The protein interactions with the superimposable benzamide rings of 18, 20, DFMBA, PC190723, and TXA6101 are conserved ( Figures 6A and S6). Nevertheless, these interactions appear insufficient for strong binding, as suggested by the 66% occupancy of DFMBA, compatible with its low affinity, while that of 18 and 20 is 100%. On the other hand, the halogenated phenyl moieties of 18 and 20 bind deep into the pocket with their benzene rings connecting to T309 and I311. This last residue adopts the same conformation in all examined complexes, while T309 acts as a gate presenting two different conformations, closed and open. The first is observed with DFMBA, the smaller bound molecule not reaching T309, and with TXA6101. In this case, the hydroxyl group of the T309 side chain points to the solvent, while the methyl group lies next to G196, thus closing the central region of the binding site ( Figure 6B, purple). The open conformation is observed in the presence of 18, 20, or PC190723. In this arrangement, the T309 side chain points its hydroxyl group toward N263 and allows access to the central region of the binding site ( Figure  6B, purple). Interestingly, compounds 18 and 20 adopt the flattest geometry, with their aromatic rings located in the same plane, as compared to PC190723 and TXA6101 ( Figure 6C). This geometry places the halogenated benzene ring close to the position occupied by the oxazole ring of the flexible compound TXA6101 (Chart S1), where the 5-bromo group interacting with a deep hydrophobic pocket was linked to the enhanced compound activity. 31 Unlike TXA6101, compounds 18, 20, and PC190723 do not displace the M226 side chain ( Figure 6C), which constitutes another gate ( Figure 6B (Table S6). Recently, the crystal structure of SaFtsZ in complex with a fluorescent BODIPY derivative of TXA6101 bound into the interdomain cleft has been reported. Nonetheless, this probe (BOFP, Chart S1) binds to unassembled SaFtsZ in solution without nucleotide and magnesium, as well as to Gram-negative FtsZs, in puzzling contrast with its complex filament crystal structure, 48 the specificity of our probes, and the characteristic selectivity of PC190723 analogues. SaFtsZ mutations conferring resistance to the inhibitors can be analyzed based on the SaFtsZ-18 and -20 complex structures. In particular, mutations G196S, M262I, and T309I confer resistance to 20. The trifluoromethyl group of 20 is placed close to G196 and the introduction of a side chain in this position would produce steric clashes similarly to PC190723. Residue M262 locates at the center of the β-sheet that forms one side of the ligand-binding cleft. While its side chain points away from the cleft, the M262I mutation can modify the hydrophobic packing of the β-sheet and, thus, potentially affect inhibitor binding. An important modification is introduced in the binding cleft with the T309I mutation. The introduction of a bulkier residue will likely impede its movement, keeping closed the central region of the binding site, thus clogging the binding of compounds.
In summary, our crystallographic analysis confirms the allosteric site for compounds 18 and 20 and provides insights into the binding to FtsZ of simplified benzamide-based bacterial cell division inhibitors. Their halogenated phenyl rings, replacing the larger chloro-thiazolopyridine moiety of PC190723, structurally superpose to the middle bromooxazole ring of TXA6101 upon binding to FtsZ. Potential ligand extensions might reach the ligand-induced cavity occupied by the TXA6101 trifluoromethyl group or the upper unfilled cavity in FtsZ.

■ CONCLUSIONS
The allosteric benzamide binding site at the cleft between the nucleotide-binding and GTPase-activating domains of FtsZ has been explored in this work. This cleft opens when FtsZ switches into its active conformation for assembly into dynamic polar filament clusters that guide bacterial division. We have developed a novel competitive assay employing specific fluorescent probes to screen for allosteric inhibitors and evaluate their binding affinity. Fluorescence polarization screening, together with cytological profiling and X-ray crystallography, has permitted us to identify effective, structurally simplified benzamide FtsZ-targeting inhibitors of bacterial division in B. subtilis and methicillin-resistant S. aureus. Our in vitro competitive binding assay and in situ cellbased methodology, combined with structural insights into ligand extension, should facilitate the discovery and validation ■ EXPERIMENTAL SECTION Synthesis. Unless stated otherwise, starting materials, reagents, and solvents were purchased as high-grade commercial products from Abcr, Acros, Scharlab, Sigma-Aldrich, or Thermo Fisher Scientific and were used without further purification. All nonaqueous reactions were performed under an argon atmosphere in oven-dried glassware. Tetrahydrofuran (THF), diethyl ether, and dichloromethane (DCM) were dried using a Pure Solv Micro 100 Liter solvent purification system. Triethylamine was dried over KOH and distilled before use. Reactions under MW irradiation were performed in a Biotage Initiator 2.5 reactor. Reactions were monitored by analytical thin-layer chromatography (TLC) on silica gel plates supplied by Merck (Kieselgel 60 F-254) with detection by UV light (254 nm) or 10% phosphomolybdic acid solution in EtOH. Flash chromatography was performed on a Varian 971-FP flash purification system using silica gel cartridges (Varian, particle size 50 μm). Melting points (mp, uncorrected) were determined on a Stuart Scientific electrothermal apparatus. Infrared (IR) spectra were measured on a Bruker Tensor 27 instrument equipped with a Specac attenuated total reflection (ATR) accessory of 5200−650 cm −1 transmission range; frequencies (ν) are expressed in cm −1 . Nuclear magnetic resonance (NMR) spectra were recorded on a Bruker Avance III 700 MHz ( 1 H, 700 MHz; 13 C, 175 MHz), Bruker Avance 500 MHz ( 1 H, 500 MHz; 13 C, 125 MHz), or Bruker DPX 300 MHz ( 1 H, 300 MHz; 13 C, 75 MHz) instruments at the Universidad Complutense de Madrid (UCM) NMR core facilities. Chemical shifts (δ) are expressed in parts per million relative to internal tetramethylsilane; coupling constants (J) are in hertz (Hz). The following abbreviations are used to describe peak patterns when appropriate: singlet (s), doublet (d), triplet (t), quartet (q), quintet (qt), sextet (sext), multiplet (m), broad (br), and apparent (app). Two-dimensional (2D) NMR experiments (heteronuclear multiple quantum coherence (HMQC) and heteronuclear multiple bond correlation (HMBC)) of representative compounds were carried out to assign protons and carbons of the new structures. High-resolution mass spectrometry (HRMS) was carried out on a Bruker FTMS APEX-Q-IV spectrometer in an electrospray ionization (ESI) mode or on a Bruker matrix-assisted laser desorption ionization time-of-flight (MALDI-TOF)/TOF ULTRAFLEX spectrometer at the UCM's mass spectrometry core facility. High-performance liquid chromatography coupled to mass spectrometry (HPLC-MS) analysis was performed using an Agilent 1200LC-MSD VL instrument. LC separation was achieved with a Zorbax Eclipse XDB-C18 column (5 μm, 4.6 mm × 150 mm) or a Zorbax SB-C3 column (5 μm, 2.1 mm × 50 mm), both together with a guard column (5 μm, 4.6 mm × 12.5 mm). The gradient mobile phases consisted of A (95:5 water/ acetonitrile) and B (5:95 water/acetonitrile) with 0.1% ammonium hydroxide and 0.1% formic acid as the solvent modifiers. MS analysis was performed with an ESI source. The capillary voltage was set to 3.0 kV and the fragmentor voltage was set at 72 eV. The drying gas temperature was 350°C, the drying gas flow was 10 L/min, and the nebulizer pressure was 20 psi. Spectra were acquired in a positive or negative ionization mode from 100 to 1200 m/z and in a UV mode at four different wavelengths (210, 230, 254, and 280 nm). Optical rotation [α] was measured on an Anton Paar MCP 100 modular circular polarimeter using a sodium lamp (λ = 589 nm) with a 1 dm path length; concentrations (c) are given as g/100 mL. The enantiomeric excess (ee) was determined by HPLC using a chiral column (Chiralpak IA, 5 μm, 4.6 mm × 150 mm) and hexane/ isopropanol/triethylamine/trifluoroacetic acid (70/30/0.3/0.1) as mobile phase (1 mL/min). HPLC traces were compared to racemic samples obtained by mixing equal amounts of the enantiopure compounds obtained independently.
Spectroscopic data of all described compounds were consistent with the proposed structures. Satisfactory HPLC-MS chromatograms were obtained for all tested compounds, which confirmed a purity of at least 95%.
General Procedure for the Synthesis of Final Compounds 2−5. To a solution of amine 35−37 or 40 (1 equiv) and Cs 2 CO 3 (3 equiv) in dry acetonitrile (50 mL/mmol), a solution of NBD-Cl (1− 1.2 equiv) in dry acetonitrile (1 mL/mmol) was added and the reaction was stirred at 80°C for 1 h. Then, the solvent was evaporated under reduced pressure and the crude was purified by chromatography (from hexane to hexane/EtOAc, 1:2) to afford final compounds 2−5. Preparation of Cross-Linked FtsZ Polymers. FtsZ (10−15 μM) was assembled in HEPES buffer, 10 mM MgCl 2 , 50 μM GMPCPP at 25°C for 10 min, and then 0.15% (v/v) glutaraldehyde 36 (distilled grade for microscopy, TAAB Laboratories, U.K.) was added to the solution that was incubated at 25°C for 10 min more. (This was the minimal glutaraldehyde concentration determined to stabilize FtsZ polymers specifically binding probe 1.) The remains of the cross-linking agent were quenched by adding 60 mM NaBH 4 , the sample was incubated on ice for 10 min and degassed. 36 Cross-linked polymers were centrifuged for 10 min at 8200g (5000 rpm) and 4°C in 15 mL Falcon tubes employing a Rotina 380R (Hettich) centrifuge, the supernatant was removed and the pellet was resuspended in the same volume of HEPES buffer, 10 mM MgCl 2 , containing 5 μM GMPCPP. Fixed FtsZ polymers were active in binding assays after more than 2 days at 4°C; they could also be frozen in liquid nitrogen with a small loss of binding capacity. However, they were observed to precipitate above 20 μM FtsZ concentration. Cross-linked polymers of SaFtsZ were similarly prepared.
Stoichiometry and Affinity of Binding of the Fluorescent Probes to FtsZ Polymers. The stoichiometry of binding of (R)-5 to stabilized FtsZ polymers was measured using a centrifugation assay adapted from ref 49. Then, FtsZ polymers (10 μM) were incubated at 25°C in HEPES buffer, 10 mM MgCl 2 , 2% DMSO, and 0.1 mM GMPCPP at 25°C in the presence of different probe concentrations, in a final volume of 0.1 mL. The samples were then centrifuged for 20 min at 100 000 rpm and 25°C in a TLA-100 rotor. After centrifugation, the supernatant was carefully withdrawn and the pellets were resuspended in the same volume of buffer. The concentration of the free probe was determined spectrophotometrically in the supernatant, employing an extinction coefficient ε 483 = 25 400 M −1 cm −1 for (R)-5 and the concentration of probe bound to FtsZ was calculated as the difference of the known total concentration of probe minus the free concentration. To calculate the FtsZ polymer concentration, 5 μL of the resuspended pellet was applied to sodium dodecyl sulfate (SDS)-12% polyacrylamide gels, stained with Coomassie blue, scanned with a CS-800 calibrated densitometer (BioRad), and the protein bands quantified using Quantity One software (BioRad).
Binding affinity of the probes to FtsZ was measured by an increase in anisotropy of the probe (adapted from ref 49). Fixed concentrations of the probes (3 μM) were titrated with different FtsZ polymer concentrations (0−40 μM) in HEPES buffer with Mg 2+ , to obtain the anisotropy increment, Δr max , corresponding to the bound probe. The increase in anisotropy was plotted against the FtsZ polymer concentration (calculated by subtracting from the total protein concentration the critical concentration for polymerization under the same conditions) and iteratively least-squares fitted with an isotherm of binding to one site. The estimated values of Δr max were used to approximate the free FtsZ concentrations, and these new values were employed again until an unchanging Δr max value was obtained. The convergent data were used to calculate the binding constant of the fluorescent probe to FtsZ polymers (Table S1).
Affinity of Ligands Competing with the Fluorescent Probes. where r free and r max are the anisotropy values of the free and bound probes, respectively, and R is the ratio between the fluorescence intensities of the bound and free probe (Table S1). The binding data were plotted against the competing ligand concentration and fitted by competition equilibrium of the ligand and the probe for the same binding site. The resulting system of equations 50 was numerically solved using the Equigra v. 5.0 program, 33 which provided the bestfitted value of the competitor binding constant. Cellular Methods. B. subtilis 168 and Staphylococcus spp. were grown in cation adjusted Mueller−Hinton broth (CAMHB; Becton, Dickinson and Company) at 37°C. B. subtilis SU570 were grown in the same medium at 30°C. In all cases, the cells were grown to an absorbance of 0.1−0.2 at 600 nm and then incubated either with the compounds or DMSO vehicle. Microscopy assays and cell measurement were performed as previously described. 31 MIC values were determined by a broth macrodilution method. 51 MBC values were determined as the lowest concentration of inhibitor at which there was a 99.9% reduction in colony-forming units/mL compared to the inoculum. To construct a S. aureus strain expressing FtsZ-mCherry, the plasmid pCN-ftsZmch 9 was electroporated into RN4220 and 0.1 μM cadmium chloride (Sigma-Aldrich) was used to induce the expression of the construct under the control of the Pcad promoter.
Isolation of FtsZ-inhibitor-resistant mutants was performed by plating 100 μL of S. aureus Mu50 culture (∼10 8 cells/mL) into CAMHB supplemented with the corresponding FtsZ inhibitor at a concentration higher than the MIC and 2% DMSO. The plates were incubated for 24 h at 37°C, and several colonies were picked for further analysis. Genomic DNA was extracted from each colony using the Wizard Genomic DNA Purification Kit (Promega) and was used in polymerase chain reactions (PCR) for amplification of f tsZ using the primers SaFtsZ-F (5′ TGGCCAATAAAACTAGGAG 3′) and SaFtsZ-R (5′ TGTTATCTGATGATTTGTGTTG 3′). The resulting PCR products were purified using the DNA gel extraction kit (Cultek) and sequenced (Macrogen Inc., Spain). Sequence analysis was performed with BLAST (NCBI).
The sensitivity of the IMR90 cell line to compounds was tested through a standard 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) assay. 52 The results were reported as the viability percentage of the tested compound relative to vehicle (DMSO), obtained from two or three independent experiments performed in triplicate.
Crystallization and Structure Determination. Crystallization assays were carried out using purified truncated SaFtsZ(12−316) at 14 mg/mL. For cocrystallization experiments, compounds (at 1−4 mM concentrations, depending on solubility) and 4% (v/v) 1-methyl-2-pyrrolidone (compound solvent; Sigma, analytical standard) were included in the protein solution. Crystals were grown at 295 K by vapor diffusion (sitting drop) under previously employed conditions: 25,34 0.2 M lithium sulfate, 10% ethylene glycol, 0.1 M Tris/ HCl, pH 8.4−9.0, and 24−28% PEG5000 MME. The crystals employed to solve the structure of the SaFtsZ-DFMBA complex (Table S5) were obtained with 4 mM DFMBA at 27% PEG5000 MME, pH 8.7. Crystal soaking experiments were typically made with 5−20 mM compound and 10−20% (v/v) 1-methyl-2-pyrrolidone for 16−20 h. For the SaFtsZ-18 and SaFtsZ-20 complex structures (Table S5), crystals grown in 28% PEG5000 MME, pH 8.6, and 25% PEG5000 MME, pH 8.4, respectively, were soaked in a 10 mM compound and 20% (v/v) 1-methyl-2-pyrrolidone. All crystals were flash-cooled by immersion in liquid nitrogen. Diffraction data were collected at the ALBA synchrotron (Spain) and the ESRF synchrotron (France). All data were processed using XDS 53 and Aimless from the CCP4 Suite. 51 Data collection and refinement statistics are presented in Table S5. The structures were determined through molecular replacement using Molrep 54 and the PDB entry 3VOA as a searching model. Model building and refinement were done using Coot 55 and PHENIX, 56 respectively. Refinement statistics are summarized in Table S5. Structural figures were prepared using PyMOL (Schrodinger Inc.).
Fluorescent benzamide probes synthesized in this work ( Figure S1); isotherms of binding of the fluorescent probes 1, (S)-2, and (R)-5 to FtsZ polymers measured by fluorescence anisotropy ( Figure S2); competition isotherms of weak FtsZ inhibitors from the binding screen ( Figure S3); effect of high affinity FtsZ inhibitors on bacterial cell division ( Figure S4); expanded view of the SaFtsZ binding cleft, with omit electron density maps (green), contoured at 3σ around compounds DFMBA, 18, and 20 ( Figure S5); ligand interaction diagram of residues located in the binding site ( Figure  S6); binding parameters of the fluorescent probes to FtsZ-GMPCPP polymers and activity on bacterial cells (Table S1); cell length of B. subtilis 168 and diameter of S. aureus cells incubated with FtsZ inhibitors (Table S2); cytotoxic effect of FtsZ inhibitors in IMR90 cell line (Table S3); spontaneous mutations of inhibitor-resistant S. aureus strains (Table S4); data collection and refinement statistics (Table S5); structure of FtsZ inhibitors TXA707, TXA709 and TXA6101, and fluorescent probe BOFP (Chart S1); structure of previously identified FtsZ inhibitors screened for binding against the interdomain cleft of the protein in this work (Chart S2); synthetic scheme S1; detailed synthetic procedures and characterization data of intermediates 31−37, 39−47, and 50; final compounds 6−17, 19, 21, 22−24 and probes SP1−SP5; solubility of all tested compounds (Table S6); and NMR spectra of final compounds and HPLC traces analysis (PDF) Molecular formula strings (CSV)